Deaerator cracking - a new problem in old equipment

TWI Bulletin, June 1987

by Richard Walker

Richard Walker, PhD, CEng, MIM, is Leader of the Stainless Steels and Corrosion Section in the Materials Department.

Widespread cracking of weld regions in deaerator vessels has become recognised as a major industrial problem only in recent years. Possible reasons for the cracking are discussed and a new Welding Institute Group Sponsored Research Project, designed to identify the cracking mechanism and provide guidelines on safe plant operation, is outlined.

Failure modes such as fatigue, creep, brittle fracture or corrosion regularly occur and are familiar to the investigative metallurgist. Such failure types have been well documented over the decades.^[1] Occasionally, however, an apparently new failure type arises which demands greater effort towards its understanding and, ultimately, its solution. Such a problem has developed recently with regard to widespread, serious cracking and, in some instances, explosive failures of deaerator vessels.

This article summarises the current state of understanding of deaerator vessel cracking. Following a brief description of deaerator operation, the recent history of the problem is outlined, and current views on the mechanism of cracking are discussed. Finally, reference is made to a Group Sponsored Research Programme being initiated at Abington with the aims of defining conditions causing cracking and of enabling guidelines to be formulated for safe plant operation and inspection procedures.

Deaerator operation

As their name implies, deaerator vessels have the function of removing air or oxygen from a process stream, which is generally, although not exclusively, a hot water supply to a steam raising, or boiler plant. The main object of deaeration is to prevent corrosion of steel plant, since even small levels of oxygen may cause significant attack. In modern, high pressure boilers, for example, residual, dissolved oxygen levels as low as 5 ppb are required.

Although several designs of deaerator are available,^[2] their operation is essentially similar in that deaeration is carried out by a counter current of steam and water within a pressurised steel vessel (Fig.1). The steam heats the feed water to boiling point and strips oxygen from the water. Operating conditions may vary between typically 1.2 and 10 bars pressure and 105-180°C, producing residual oxygen levels ranging from a low of 5 ppb to a maximum of typically 1 ppm. Under such operating conditions, a thin film of magnetite (Fe₃O₄) is formed on steel surfaces and, since this film is normally protective, no significant corrosion failures are expected either within the carbon steel deaerator vessel or down stream in the associated plant system.

Deaerator cracking

The background

In 1983, a major plant failure occurred in Pine Hill, Alabama, when complete, explosive rupture of a deaerator vessel took place in a paper mill. This serious incident prompted the Technical Association of the Pulp and Paper Industry (TAPPI) in North America to examine urgently the condition of other deaerators in the industry. Astonishingly, at least 68 other deaerator vessels, representing over 50% of 1983 inspections, were found to be cracked in weld regions. As a consequence, a TAPPI Engineering Division advisory notice was issued in September 1983 to alert operators of the potential dangers of deaerator failures and the need for urgent inspections to be carried out. ^[3]

Following the TAPPI initiative, the National Association of Corrosion Engineers (NACE), Houston, set up Committee T7H7, in April 1984, to study weld cracking and rupture of boiler feedwater deaerators.^[4] With the involvement of NACE, a wider cross section of both European and North American industries, including petrochemical, food processing and, in particular, electrical power generation representatives became concerned with the problem.^[5-9] It became evident that the initial findings reported by TAPPI were not exclusive to the pulp and paper industry and that some 30-40% of deaerators across a range of industries were cracked.

The widespread nature of the problem is not thought to reflect a sudden, coincidental deterioration in deaerator plant worldwide, but rather the increased vigilance of operators and inspection authorities and the clear attention paid to vessels which previously might have had only limited visual or non-destructive examination. Paradoxically, it is the usually protective magnetite film which builds up during normal deaerator operation which may hide the presence of cracks. As a consequence, the need for careful non-destructive testing especially by magnetic particle inspection has been stressed. ^[3,10]

Nature of cracking

Deaerator problems have usually taken the form of cracking of both longitudinal and circumferential weld areas,^[10] although it is possible that circumferential (girth) joints, especially dished end welds, may be more prone to failure.^[11,12] Cracks have occurred transverse and parallel to welds, both in the weld metal and heat affected zone (HAZ) areas. In addition, cracking in the parent plate has been observed opposite external attachment welds. In one case, a lifting lug welded on the outside of a deaerator was found to be associated with internal cracking penetrating up to two thirds of the wall thickness after 12 years of service.^[4] Cracking has been found primarily in drums holding deaerated water below the waterline although the water level can vary considerably depending on plant design and loading requirements.

While extensive weld associated cracking may be present,^[6] it is generally reported to be of shallow depth (e.g. 1-3mm) and, since such typical defects have frequently been observed in old plant,^[5] the rate of crack growth is often considered to be low, although, as discussed below, this will depend to some extent on the precise mechanism of cracking. The presence of weld cracking especially in older plant may in some cases reflect original manufacturing defects, rather than a crack growth mechanism during service.^[9] Nonetheless there is sufficient evidence to suggest that major cracks found in deaerator vessels have grown during the lifetime of the plant.

In general, deaerator cracking shows the characteristic form of being straight and transgranular in nature, roughly normal to the steel surface, with little branching evident. The cracks are commonly blunt ended, show evidence of corrosion and are tightly filled with an iron oxide corrosion product (magnetite). Figure 2 shows a longitudinal section through a double V butt weld in an 18mm wall deaerator vessel. The transverse weld metal cracking shown penetrated 85% of the vessel wall over 19 years during which 3700 plant start-ups were recorded. The low quality of welding in this old deaerator is clearly apparent. The crack detail (Fig.3 and 4) illustrates the transgranular nature of the cracking and the continuous oxide film within the crack.

Causes of cracking

In surveys of deaerators carried out by TAPPI and NACE and other organisations, a large number of incidents of failure have been examined in detail. Deaerators have been fabricated from a wide range of materials and in a variety of designs, yet there is no apparent correlation between cracking and steel type or fabrication procedure.^[13] Similarly, no correlations have been observed between incidents of failure and the age, size, operating pressure or type of water treatment. ^[14]

In some cases deaerator cracking has been readily explained. For example, the carry over of caustic salts from boiler water return has led to stress corrosion cracking,^[4] with its characteristic intergranular crack path. In other cases, cracking has been attributed to thermal fatigue effect.^[15] There is also no doubt that original manufacturing defects account for some of the incidents of cracking. ^[9]

However, the vast majority of deaerator failures, which do not fall into the above categories, are considered to result from environmentally assisted cracking.^[4,13] Two main failure mechanisms have been proposed, namely corrosion fatigue and 'stress induced corrosion'. In addition, hydrogen embrittlement of fine grained, high strength steels has been suggested by German investigators as a third possible cause of cracking.^[16] These workers, investigating deaerator vessel failures some years prior to the American experiences, found differences in the mode of cracking between carbon steel and high strength steel materials. In the latter case, cracks were much finer, sometimes with clear signs of hydrogen induced cold cracking at the origin, and, although no precise mechanism for crack propagation was determined, it was considered that hydrogen embrittlement may have contributed to failure. In deaerator service, atomic hydrogen can be liberated by the net reaction between steel and hot water to form magnetite (Fe₃O₄):

₃Fe + ₄H₂O → Fe₃O₄ + 8H     ...[1]

Hydrogen cracking might be a contributory factor towards failure, but deaerator vessels, in general, are not made from high strength steels while cracking has occurred in weld areas sufficiently soft for their sensitivity to hydrogen embrittlement to be low. The most usually accepted cause of cracking, therefore, is the conjoint action between stress and corrosion, which results in corrosion fatigue or stress induced corrosion mechanisms of failure.

In corrosion fatigue, crack initiation and propagation are related to the simultaneous action of cyclic stresses and corrosion, as may arise from thermal or mechanical variations in the system.^[13] Although largely protective when formed at higher temperatures, magnetite formation below about 230°C tends to produce a more defective and thinner surface oxide coating^[17] which is susceptible to brittle failure under cyclic stress. Under fatigue conditions, it is argued that fracture of the magnetite layer exposes the underlying metal to corrosion [I], and new magnetite is formed, producing a localised notch. Subsequent stress cycling causes fracture of the oxide layer at the notch causing it to deepen. As the process continues a wedge shaped, oxide filled crack propagates. ^[18]

Fluctuating stresses within a deaerator vessel may arise from a number of sources, and a major contribution is reported to be water hammer in the inlet system, caused for example by the mixing of hot condensate with a cold make up stream. One survey^[19] reports that of 18 deaerators known to suffer water hammer in service, 14 were found to be cracked. Of the four not cracked, three had been in service less than four years. Similarly, of 20 deaerators known to be free from water hammer, only three were found to be cracked and these were installed in the 1960s or in one case as long ago as 1946.

It is evident that, in the case of a corrosion fatigue mechanism of failure, crack growth will occur during operation under the influence of the simultaneous action of corrosion and fluctuating stress. In general, therefore, a relatively slow crack growth rate may be assumed. With a stress induced corrosion mechanism of failure, however, the variable stress and corrosion components of the mechanism may not necessarily act concurrently. The process of stress induced corrosion is considered to occur in boiler systems where localised stress induced damage to a protective magnetite film in service may be followed by corrosion during periods out of operation.^[17] It is possible that damage to the magnetite layer takes place during the process of shutting down the plant, following which corrosion of the underlying metal surface would occur. Furthermore, under the higher oxygen levels which exist during a plant shut down, significant localised corrosion would be possible because of a large galvanic corrosion effect associated with the unfavourable area ratio between the cathodic magnetite scale and the small anodic region of exposed metal. In such a case, corrosion would be related to the cathodic reduction of O₂ rather than the direct reaction between iron and water shown in [1].

It is clear that, with a stress induced corrosion mechanism of failure, the major amount of crack growth would occur during or close to plant shut down periods and, as such, the actual crack growth rates would be significantly higher than in the corrosion fatigue mechanism, although cracking would develop only for a restricted time.

At present, it is not clear which of the mechanisms of failure is the more important and it is possible that both may contribute to deaerator cracking. In either case, an important consideration is the level of protection afforded by the magnetite scale and there are a number of reasons why this may be reduced in weld areas: the presence of residual welding stresses, weld bead surface roughness and the application of surface grinding or machining to weld preparation areas, may all result in the formation of weld area magnetite layers which are less protective than on parent steel, and hence account for the widespread occurrence of cracking in weld regions. Certainly residual welding stresses must play a significant role in the cracking mechanism, since it is known that the risk of deaerator cracking can be reduced by carrying out a post-weld stress relief procedure. ^[4]

Present position

Practical implications and the need for research

It is evident that the recently recognised problem of deaerator cracking is not so much caused by a new failure mechanism but rather by a variation on an old theme - namely environmentally assisted cracking. Both corrosion fatigue or stress induced corrosion may contribute to failure, mechanisms largely thought to be associated with on-stream or off- stream crack growth respectively. There appears to be little support for a hydrogen embrittlement mechanism of failure. ^[4]

As discussed above, the distinction between corrosion fatigue and stress induced corrosion is an important one, since these failure mechanisms imply quite different crack growth rates.

With the knowledge gained in the interim since the TAPPI advisory note was issued, industry has had little choice but to inspect its deaerators regularly and remove cracks when found (with or without subsequent repair welding). Such an inspection and repair programme is extremely costly and could be greatly reduced if more precise data were available on crack growth mechanisms and, in particular, crack growth rates. With these points in mind The Welding Institute has prepared a Group Sponsored Research Project with the following objectives; ^[20]

1. To identify conditions under which weld area cracking in deaerator vessels is likely to develop;
2. To provide guidelines on methods of avoiding failure in operating vessels and equipment;
3. To generate data on crack growth rates so that a) plant inspections are not unnecessarily frequent and b) the need for immediate removal/rectification of identified cracks can be clearly defined.

The test programme will include both atmospheric and higher pressure (autoclave) boiling water conditions, simulating deaerator environments of a range of O₂ levels, pH, temperature and pressure. Both dynamic and static loading of deaerator plate and weld samples will be carried out, and environmental and material conditions identified as particularly severe will be used during measurement of crack growth rates. On completion of the research programme, the conditions conducive to weld area cracking in deaerators will have been better defined and guidelines formulated to establish realistic frequency of inspection.

The research programme started in April 1987 and further information may be obtained from the author or from the project leader David Sparkes.

Acknowledgements

The author thanks his colleagues for advice and assistance in the production of this article, and the Member company concerned for permission to use Fig.2-4.

References

1. Case histories in failure analysis, American Society for Metals, 1979.
2. Solt G S: 'The practical design of heater deaerators'. Dewplan Technical Bulletin TB16, DEWPLAN (WT) Limited, High Wycombe, Bucks, 1986.
3. Technical Association of the Pulp and Paper Industry: 'First advisory note CA4689/CA4690'. TAPPI Engineering Division - Deaerator Advisory, 30 September 1983, Atlanta, Georgia.
4. McIntyre D R: 'A review of corrosion and cracking mechanisms in boiler feed deaerators'. Paper 309, Corrosion 86 1986 17-21 March Houston, NACE.
5. National Board Bulletin: 'More on those dangerous deaerators'. The National Board of Boiler and Pressure Vessel Inspectors, Columbus, Ohio. 1984 July.
6. Deaerator cracking problems - panel discussion. UK Corrosion 85, 1985 6 November Harrogate I Corr ST/NACE.
7. Deaerator cracking - panel discussion. Corrosion 86 17-21 March Houston NACE.
8. Pastoors J T W: 'Deaerator cracking in Dutch utility boilers'. Presented at VGB Kraftwerkskomponenten Essen 1986 5-6 March.
9. Gooch T G and Hart P H M: 'Review of welding practice for carbon steel deaerator vessels'. Materials Performance 1986 December 30-38.
10. Allen C: 'Magnetic particle testing pinpoints cracks in deaerator vessels'. Power 1985 129 January 55-57.
11. Private communication, Welding Institute Research Member company.
12. 'Deaerator weld cracking - a major problem'. Esso Engineering Panel Discussion, UK Corrosion 85, 1985 6 November, Harrogate.
13. Chakrapani D G: 'Environmentally-assisted cracking in welded deaerator vessels in the pulp and paper industry'. Paper 138 Corrosion 86 17-21 March Houston, NACE.
14. Clevenger T G: 'Deaerator cracking - an industry update'. Ibid Paper 302.
15. O'Keefe P: 'The role of metallography and NDT in the analysis of a failed steam accumulator'. Materials Evaluation 1978 36 (9) 40-45.
16. Adamsky F J and Teichmann H D: 'Operating experience with feed water tanks'. VGB Kraftwerkstechnik 1977 57 November 730-743.
17. Schoch W and Spahn H: 'On the role of stress induced corrosion and corrosion fatigue in the formation of cracks in water wetted boiler components'. Proc conf on 'Corrosion fatigue: chemistrymechanisms and microstructure', Univ of Connecticut. 1971 14-18 June NACE 52-64.
18. Kelly J A: 'Operation and water chemistry in deaerator cracking'. Paper 304 Corrosion 86 1986 17-21 March Houston NACE.
19. Robinson J O: 'Deaerator cracking survey: basic design, operating history, and water chemistry survey'. Ibid Paper 305.
20. The Welding Institute Contract Proposal 'Cracking of weldments in feed water deaerator systems'. CP/MAT/2690-2 1986 November.